---
title: "Tableau de Bord du Cycle de Vie Client E-commerce"
output: 
  flexdashboard::flex_dashboard:
    theme: united
    orientation: row
    vertical_layout: fill
    social: ["twitter", "facebook", "linkedin"]
    source_code: embed
    navbar:
      - { title: "Télécharger le rapport", icon: "fa-file-pdf", href: "#", align: right }
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, warning = FALSE, message = FALSE)

# Chargement des packages
library(flexdashboard)
library(tidyverse)
library(readxl)
library(plotly)
library(DT)
library(lubridate)
library(cluster)
library(factoextra)
library(RColorBrewer)
library(scales)
library(prophet)
library(treemap)
library(arules)
library(arulesViz)
library(htmltools)
library(htmlwidgets)
library(shiny)
library(fontawesome)
```

```{r data-load, include=FALSE}
# Chargement et préparation des données
online_retail <- read_excel("Online_Retail.xlsx")

# Nettoyage des données
online_retail_clean <- online_retail %>%
  filter(!is.na(CustomerID)) %>%
  filter(Quantity > 0, UnitPrice > 0) %>%
  mutate(
    InvoiceDate = as.Date(InvoiceDate),
    TotalPrice = Quantity * UnitPrice,
    Year = year(InvoiceDate),
    Month = month(InvoiceDate, label = TRUE),
    Quarter = quarter(InvoiceDate),
    Weekday = wday(InvoiceDate, label = TRUE),
    YearMonth = floor_date(InvoiceDate, "month")
  )

# Date de référence (un jour après la dernière date)
max_date <- max(online_retail_clean$InvoiceDate) + 1

# Calculer les métriques RFM
rfm_data <- online_retail_clean %>%
  group_by(CustomerID) %>%
  summarise(
    Recency = as.numeric(max_date - max(InvoiceDate)),
    Frequency = n_distinct(InvoiceNo),
    Monetary = sum(TotalPrice),
    FirstPurchase = min(InvoiceDate),
    LastPurchase = max(InvoiceDate),
    AvgOrderValue = Monetary / Frequency,
    PurchaseSpan = as.numeric(max(InvoiceDate) - min(InvoiceDate)),
    PurchaseFrequency = ifelse(PurchaseSpan > 0, Frequency / (PurchaseSpan / 30), 0)
  ) %>%
  mutate(
    CustomerAge = as.numeric(max_date - FirstPurchase),
    DaysSinceLastPurchase = Recency,
    MonthlySpend = ifelse(CustomerAge > 0, (Monetary / CustomerAge) * 30, 0)
  )

# Normalisation et clustering
rfm_normalized <- rfm_data %>%
  mutate(
    RecencyNorm = scale(Recency),
    FrequencyNorm = scale(Frequency),
    MonetaryNorm = scale(Monetary)
  )

rfm_for_clustering <- rfm_normalized %>%
  select(RecencyNorm, FrequencyNorm, MonetaryNorm)

# Appliquer K-means avec 5 clusters
set.seed(123)
k <- 5
kmeans_result <- kmeans(rfm_for_clustering, centers = k, nstart = 25)

# Ajouter les clusters aux données RFM
rfm_with_clusters <- rfm_normalized %>%
  mutate(Cluster = as.factor(kmeans_result$cluster))

# Attribuer des noms aux segments
segment_names <- c(
  "1" = "Champions",
  "2" = "Clients Fidèles",
  "3" = "Clients Potentiels",
  "4" = "Clients à Risque",
  "5" = "Clients Dormants"
)

# Ajouter les noms des segments
rfm_with_segments <- rfm_with_clusters %>%
  mutate(Segment = segment_names[as.character(Cluster)])

# Analyses supplémentaires
# 1. Churn risk
churn_threshold <- 90
rfm_with_churn <- rfm_with_segments %>%
  mutate(
    ChurnRisk = case_when(
      Recency > churn_threshold & Frequency <= 3 ~ "Élevé",
      Recency > churn_threshold / 2 & Frequency <= 5 ~ "Moyen",
      Recency > churn_threshold / 3 ~ "Faible",
      TRUE ~ "Très faible"
    )
  )

# 2. Customer lifecycle stage
rfm_with_lifecycle <- rfm_with_churn %>%
  mutate(
    LifecycleStage = case_when(
      CustomerAge <= 30 & Frequency <= 2 ~ "Nouveau",
      Segment %in% c("Champions", "Clients Fidèles") ~ "Établi",
      Segment == "Clients Potentiels" ~ "Développement",
      Segment == "Clients à Risque" ~ "En danger",
      Segment == "Clients Dormants" ~ "Inactif",
      TRUE ~ "Autre"
    )
  )

# 3. Calcul des agrégats pour les KPIs
total_customers <- n_distinct(rfm_with_lifecycle$CustomerID)
total_revenue <- sum(rfm_with_lifecycle$Monetary)
average_order_value <- mean(rfm_with_lifecycle$AvgOrderValue)
average_purchase_frequency <- mean(rfm_with_lifecycle$PurchaseFrequency)

# 4. Calcul du taux de rétention estimé par segment
retention_rates <- rfm_with_lifecycle %>%
  group_by(Segment) %>%
  summarise(
    RetentionRate = sum(Recency <= 60) / n() * 100
  )

# 5. Premier achat vs. achat répété
first_vs_repeat <- online_retail_clean %>%
  group_by(CustomerID) %>%
  arrange(InvoiceDate) %>%
  mutate(
    PurchaseNumber = row_number(),
    PurchaseType = ifelse(PurchaseNumber == 1, "Premier achat", "Achat répété")
  ) %>%
  group_by(PurchaseType) %>%
  summarise(
    TotalRevenue = sum(TotalPrice),
    AvgOrderValue = mean(TotalPrice),
    TransactionCount = n()
  )

# 6. Analyses des produits
product_analysis <- online_retail_clean %>%
  group_by(StockCode, Description) %>%
  summarise(
    TotalSales = sum(TotalPrice),
    Quantity = sum(Quantity),
    TransactionCount = n_distinct(InvoiceNo),
    CustomerCount = n_distinct(CustomerID)
  ) %>%
  arrange(desc(TotalSales))

# 7. Pays par CA
country_sales <- online_retail_clean %>%
  group_by(Country) %>%
  summarise(
    TotalSales = sum(TotalPrice),
    CustomerCount = n_distinct(CustomerID),
    TransactionCount = n_distinct(InvoiceNo)
  ) %>%
  arrange(desc(TotalSales))

# 8. Ventes temporelles
time_sales <- online_retail_clean %>%
  group_by(YearMonth) %>%
  summarise(
    TotalSales = sum(TotalPrice),
    TransactionCount = n_distinct(InvoiceNo),
    CustomerCount = n_distinct(CustomerID)
  ) %>%
  arrange(YearMonth)

# 9. Modèle de prévision simple
prophet_data <- time_sales %>%
  rename(ds = YearMonth, y = TotalSales)

model <- prophet(prophet_data)
future <- make_future_dataframe(model, periods = 6, freq = "month")
forecast <- predict(model, future)

forecast_data <- bind_rows(
  data.frame(
    Date = prophet_data$ds,
    Sales = prophet_data$y,
    Type = "Historique"
  ),
  data.frame(
    Date = tail(forecast$ds, 6),
    Sales = tail(forecast$yhat, 6),
    Type = "Prévision"
  )
)
```

Vue d'ensemble {data-icon="fa-dashboard"}
=======================================================================

Row {data-height=150}
-----------------------------------------------------------------------

### Clients totaux
```{r}
valueBox(
  formatC(total_customers, format = "d", big.mark = " "),
  caption = "Clients uniques",
  icon = "fa-users",
  color = "info"
)
```

### Chiffre d'affaires
```{r}
valueBox(
  paste0(formatC(round(total_revenue/1000000, 2), format = "f", digits = 2, big.mark = " "), " M €"),
  caption = "Chiffre d'affaires total",
  icon = "fa-euro-sign",
  color = "success"
)
```

### Valeur panier moyen
```{r}
valueBox(
  paste0(formatC(round(average_order_value, 2), format = "f", digits = 2, big.mark = " "), " €"),
  caption = "Valeur panier moyen",
  icon = "fa-shopping-cart",
  color = "primary"
)
```

### Fréquence d'achat
```{r}
valueBox(
  formatC(round(average_purchase_frequency, 2), format = "f", digits = 2),
  caption = "Achats mensuels moyens",
  icon = "fa-sync",
  color = "warning"
)
```

Row {data-height=400}
-----------------------------------------------------------------------

### Évolution du chiffre d'affaires et prévisions
```{r}
plot_ly(forecast_data, x = ~Date) %>%
  add_trace(
    y = ~Sales,
    color = ~Type,
    type = 'scatter',
    mode = 'lines',
    line = list(width = 3),
    colors = c("Historique" = "#1F77B4", "Prévision" = "#FF7F0E")
  ) %>%
  layout(
    title = "Évolution du chiffre d'affaires et prévisions à 6 mois",
    yaxis = list(title = "Chiffre d'affaires (€)", tickformat = ",.0f"),
    xaxis = list(title = ""),
    legend = list(orientation = "h", y = -0.2)
  )
```

### Distribution des segments clients
```{r}
segment_distribution <- rfm_with_lifecycle %>%
  count(Segment) %>%
  mutate(Percentage = n / sum(n) * 100) %>%
  arrange(desc(n))

plot_ly(segment_distribution, labels = ~Segment, values = ~n, type = "pie",
        textposition = "inside",
        textinfo = "label+percent",
        insidetextfont = list(color = "#FFFFFF"),
        marker = list(colors = brewer.pal(5, "Set1"),
                      line = list(color = "#FFFFFF", width = 1)),
        showlegend = FALSE) %>%
  layout(
    title = "Répartition des clients par segment",
    annotations = list(
      text = paste("Total:", formatC(total_customers, format = "d", big.mark = " ")),
      showarrow = FALSE,
      x = 0.5,
      y = 0.5
    )
  )
```

Row
-----------------------------------------------------------------------

### Contribution au chiffre d'affaires par segment
```{r}
revenue_by_segment <- rfm_with_lifecycle %>%
  group_by(Segment) %>%
  summarise(
    TotalRevenue = sum(Monetary),
    ClientCount = n(),
    AvgRevenue = TotalRevenue / ClientCount
  ) %>%
  mutate(
    Percentage = TotalRevenue / sum(TotalRevenue) * 100,
    Segment = factor(Segment, levels = Segment[order(TotalRevenue, decreasing = TRUE)])
  )

plot_ly(revenue_by_segment, x = ~Segment, y = ~TotalRevenue, type = "bar",
        marker = list(color = brewer.pal(5, "Set1"))) %>%
  layout(
    title = "Contribution au chiffre d'affaires par segment",
    xaxis = list(title = ""),
    yaxis = list(title = "Chiffre d'affaires (€)", tickformat = ",.0f"),
    annotations = list(
      x = revenue_by_segment$Segment,
      y = revenue_by_segment$TotalRevenue,
      text = paste0(round(revenue_by_segment$Percentage, 1), "%"),
      showarrow = FALSE,
      yanchor = "bottom",
      yshift = 10
    )
  )
```

### Taux de rétention estimé par segment
```{r}
plot_ly(retention_rates, x = ~Segment, y = ~RetentionRate, type = "bar",
        marker = list(color = brewer.pal(5, "Set1"))) %>%
  layout(
    title = "Taux de rétention estimé par segment (60 jours)",
    xaxis = list(title = ""),
    yaxis = list(title = "Taux de rétention (%)", range = c(0, 100)),
    annotations = list(
      x = retention_rates$Segment,
      y = retention_rates$RetentionRate,
      text = paste0(round(retention_rates$RetentionRate, 1), "%"),
      showarrow = FALSE,
      yanchor = "bottom",
      yshift = 10
    )
  )
```

Acquisition & Comportement {data-icon="fa-user-plus"}
=======================================================================

Row {data-height=100}
-----------------------------------------------------------------------

### {.valueBox}
```{r}
# Calcul du nombre de nouveaux clients au cours des 30 derniers jours
new_customers_30d <- rfm_with_lifecycle %>%
  filter(CustomerAge <= 30) %>%
  nrow()

valueBox(
  formatC(new_customers_30d, format = "d", big.mark = " "),
  caption = "Nouveaux clients (30j)",
  icon = "fa-user-plus",
  color = "success"
)
```

### {.valueBox}
```{r}
# Calcul de la valeur des premiers achats vs achats répétés
first_purchase_value <- first_vs_repeat %>%
  filter(PurchaseType == "Premier achat") %>%
  pull(AvgOrderValue)

valueBox(
  paste0(formatC(round(first_purchase_value, 2), format = "f", digits = 2, big.mark = " "), " €"),
  caption = "Valeur moyenne premier achat",
  icon = "fa-shopping-bag",
  color = "info"
)
```

### {.valueBox}
```{r}
# Valeur moyenne des achats répétés
repeat_purchase_value <- first_vs_repeat %>%
  filter(PurchaseType == "Achat répété") %>%
  pull(AvgOrderValue)

valueBox(
  paste0(formatC(round(repeat_purchase_value, 2), format = "f", digits = 2, big.mark = " "), " €"),
  caption = "Valeur moyenne achat répété",
  icon = "fa-redo",
  color = "primary"
)
```

### {.valueBox}
```{r}
# Ratio achats répétés vs premiers achats
repeat_ratio <- repeat_purchase_value / first_purchase_value

valueBox(
  formatC(round(repeat_ratio, 2), format = "f", digits = 2),
  caption = "Ratio valeur répété/premier",
  icon = "fa-chart-line",
  color = ifelse(repeat_ratio > 1, "success", "danger")
)
```

Row {data-height=400}
-----------------------------------------------------------------------

### Top 10 des pays par acquisition
```{r}
top_countries_acquisition <- country_sales %>%
  arrange(desc(CustomerCount)) %>%
  head(10)

plot_ly(top_countries_acquisition, x = ~reorder(Country, CustomerCount), y = ~CustomerCount, type = "bar",
        marker = list(color = colorRampPalette(brewer.pal(9, "Blues"))(10))) %>%
  layout(
    title = "Top 10 des pays par nombre de clients",
    xaxis = list(title = ""),
    yaxis = list(title = "Nombre de clients")
  )
```

### Produits les plus populaires pour les nouveaux clients
```{r}
# Identifier les nouveaux clients (premier achat dans les 30 derniers jours)
new_customers <- rfm_with_lifecycle %>%
  filter(CustomerAge <= 30) %>%
  pull(CustomerID)

# Trouver les produits les plus achetés par les nouveaux clients
new_customer_products <- online_retail_clean %>%
  filter(CustomerID %in% new_customers) %>%
  group_by(StockCode, Description) %>%
  summarise(
    ClientCount = n_distinct(CustomerID),
    TotalQuantity = sum(Quantity),
    TotalRevenue = sum(TotalPrice)
  ) %>%
  arrange(desc(ClientCount)) %>%
  head(10)

plot_ly(new_customer_products, x = ~ClientCount, y = ~reorder(substr(Description, 1, 30), ClientCount), 
        type = "bar", orientation = "h",
        marker = list(color = colorRampPalette(brewer.pal(9, "Oranges"))(10))) %>%
  layout(
    title = "Top 10 des produits achetés par les nouveaux clients",
    xaxis = list(title = "Nombre de nouveaux clients"),
    yaxis = list(title = "")
  )
```

Row {data-height=500}
-----------------------------------------------------------------------

### Tendances d'acquisition par mois
```{r}
# Calculer le nombre de nouveaux clients par mois
acquisition_trend <- online_retail_clean %>%
  group_by(CustomerID) %>%
  summarise(FirstPurchaseDate = min(InvoiceDate)) %>%
  mutate(FirstPurchaseMonth = floor_date(FirstPurchaseDate, "month")) %>%
  count(FirstPurchaseMonth) %>%
  rename(NewCustomers = n)

# Combiner avec les ventes mensuelles
monthly_acquisition <- time_sales %>%
  select(YearMonth, TotalSales) %>%
  left_join(acquisition_trend, by = c("YearMonth" = "FirstPurchaseMonth")) %>%
  replace_na(list(NewCustomers = 0))

plot_ly(monthly_acquisition) %>%
  add_trace(
    x = ~YearMonth,
    y = ~NewCustomers,
    type = "scatter",
    mode = "lines+markers",
    name = "Nouveaux clients",
    line = list(color = "#1F77B4", width = 3),
    marker = list(color = "#1F77B4", size = 8)
  ) %>%
  add_trace(
    x = ~YearMonth,
    y = ~TotalSales / 1000,
    type = "scatter",
    mode = "lines",
    name = "CA (milliers €)",
    yaxis = "y2",
    line = list(color = "#FF7F0E", width = 2, dash = "dash")
  ) %>%
  layout(
    title = "Tendance d'acquisition client vs. chiffre d'affaires",
    xaxis = list(title = ""),
    yaxis = list(title = "Nouveaux clients"),
    yaxis2 = list(
      title = "CA (milliers €)",
      overlaying = "y",
      side = "right"
    ),
    legend = list(orientation = "h", y = -0.2)
  )
```

### Comportement d'achat par segment
```{r}
# Calculer le comportement d'achat moyen par segment
purchase_behavior <- rfm_with_lifecycle %>%
  group_by(Segment) %>%
  summarise(
    AvgFrequency = mean(Frequency),
    AvgRecency = mean(Recency),
    AvgOrderValue = mean(AvgOrderValue),
    AvgMonthlySpend = mean(MonthlySpend)
  ) %>%
  gather(key = "Metric", value = "Value", -Segment) %>%
  mutate(
    Metric = factor(Metric, 
                   levels = c("AvgFrequency", "AvgRecency", "AvgOrderValue", "AvgMonthlySpend"),
                   labels = c("Fréquence moyenne", "Récence moyenne (jours)", "Valeur panier moyen (€)", "Dépense mensuelle (€)"))
  )

plot_ly(purchase_behavior, x = ~Segment, y = ~Value, color = ~Segment, type = "bar",
        colors = brewer.pal(5, "Set1")) %>%
  layout(
    title = "Comportement d'achat par segment",
    xaxis = list(title = ""),
    yaxis = list(title = "Valeur"),
    barmode = "group",
    facet_row = ~Metric
  )
```

Cycle de Vie & Segmentation {data-icon="fa-life-ring"}
=======================================================================

Row {data-height=100}
-----------------------------------------------------------------------

### {.valueBox}
```{r}
# Calculer le taux de clients actifs (achat dans les 90 derniers jours)
active_rate <- sum(rfm_with_lifecycle$Recency <= 90) / nrow(rfm_with_lifecycle) * 100

valueBox(
  paste0(formatC(round(active_rate, 1), format = "f", digits = 1), "%"),
  caption = "Taux de clients actifs (90j)",
  icon = "fa-heartbeat",
  color = ifelse(active_rate > 50, "success", "warning")
)
```

### {.valueBox}
```{r}
# Calculer le taux de clients à risque
at_risk_rate <- sum(rfm_with_lifecycle$ChurnRisk %in% c("Élevé", "Moyen")) / nrow(rfm_with_lifecycle) * 100

valueBox(
  paste0(formatC(round(at_risk_rate, 1), format = "f", digits = 1), "%"),
  caption = "Clients à risque de churn",
  icon = "fa-exclamation-triangle",
  color = ifelse(at_risk_rate < 30, "success", "danger")
)
```

### {.valueBox}
```{r}
# Calculer le taux de clients champions
champions_rate <- sum(rfm_with_lifecycle$Segment == "Champions") / nrow(rfm_with_lifecycle) * 100

valueBox(
  paste0(formatC(round(champions_rate, 1), format = "f", digits = 1), "%"),
  caption = "Taux de clients Champions",
  icon = "fa-trophy",
  color = "success"
)
```

### {.valueBox}
```{r}
# Calculer la valeur client moyenne
avg_customer_value <- mean(rfm_with_lifecycle$Monetary)

valueBox(
  paste0(formatC(round(avg_customer_value, 2), format = "f", digits = 2, big.mark = " "), " €"),
  caption = "Valeur client moyenne",
  icon = "fa-user-tag",
  color = "info"
)
```

Row {data-height=500}
-----------------------------------------------------------------------

### Visualisation 3D des segments clients
```{r}
plot_ly(rfm_with_lifecycle, 
        x = ~RecencyNorm, 
        y = ~FrequencyNorm, 
        z = ~MonetaryNorm, 
        color = ~Segment,
        colors = brewer.pal(5, "Set1"),
        type = "scatter3d",
        mode = "markers",
        marker = list(size = 4, opacity = 0.8)) %>%
  layout(
    title = "Segmentation clients 3D (RFM)",
    scene = list(
      xaxis = list(title = "Récence (normalisée)"),
      yaxis = list(title = "Fréquence (normalisée)"),
      zaxis = list(title = "Montant (normalisé)")
    )
  )
```

### Matrice RFM des segments
```{r}
# Créer une matrice RFM des segments (Récence vs Fréquence, taille = Montant)
rfm_matrix <- rfm_with_lifecycle %>%
  mutate(
    RecencyGroup = cut(Recency, breaks = c(0, 30, 90, 180, 365, Inf), 
                     labels = c("< 30j", "30-90j", "90-180j", "180-365j", "> 365j")),
    FrequencyGroup = cut(Frequency, breaks = c(0, 1, 3, 5, 10, Inf),
                        labels = c("1", "2-3", "4-5", "6-10", "> 10"))
  ) %>%
  group_by(RecencyGroup, FrequencyGroup, Segment) %>%
  summarise(
    ClientCount = n(),
    AvgMonetary = mean(Monetary)
  )

# Créer une matrice de chaleur
plot_ly(rfm_matrix, 
        x = ~FrequencyGroup, 
        y = ~RecencyGroup, 
        z = ~ClientCount,
        color = ~Segment,
        colors = brewer.pal(5, "Set1"),
        size = ~AvgMonetary,
        sizes = c(10, 50),
        type = "scatter",
        mode = "markers",
        marker = list(opacity = 0.8, line = list(width = 1, color = "#FFFFFF"))) %>%
  layout(
    title = "Matrice RFM des segments clients",
    xaxis = list(title = "Fréquence d'achat"),
    yaxis = list(title = "Récence")
  )
```

Row {data-height=400}
-----------------------------------------------------------------------

### Cycle de vie des clients
```{r}
# Distribution des clients par phase du cycle de vie
lifecycle_distribution <- rfm_with_lifecycle %>%
  count(LifecycleStage) %>%
  mutate(
    Percentage = n / sum(n) * 100,
    LifecycleStage = factor(LifecycleStage, 
                          levels = c("Nouveau", "Développement", "Établi", "En danger", "Inactif"))
  ) %>%
  arrange(LifecycleStage)

plot_ly(lifecycle_distribution, 
        x = ~LifecycleStage, 
        y = ~n, 
        type = "bar",
        marker = list(color = colorRampPalette(c("#4CAF50", "#FFC107", "#F44336"))(5))) %>%
  layout(
    title = "Distribution des clients par phase du cycle de vie",
    xaxis = list(title = ""),
    yaxis = list(title = "Nombre de clients"),
    annotations = list(
      x = lifecycle_distribution$LifecycleStage,
      y = lifecycle_distribution$n,
      text = paste0(round(lifecycle_distribution$Percentage, 1), "%"),
      showarrow = FALSE,
      yanchor = "bottom",
      yshift = 10
    )
  )
```

### Risque de churning par segment
```{r}
# Distribution du risque de churning par segment
churn_risk_by_segment <- rfm_with_lifecycle %>%
  group_by(Segment, ChurnRisk) %>%
  summarise(Count = n()) %>%
  group_by(Segment) %>%
  mutate(Percentage = Count / sum(Count) * 100) %>%
  arrange(Segment, desc(ChurnRisk))

# Réordonner les facteurs pour l'affichage
churn_risk_by_segment$ChurnRisk <- factor(churn_risk_by_segment$ChurnRisk, 
                                        levels = c("Très faible", "Faible", "Moyen", "Élevé"))

plot_ly(churn_risk_by_segment, 
        x = ~Segment, 
        y = ~Percentage, 
        color = ~ChurnRisk,
        colors = c("Élevé" = "#D32F2F", "Moyen" = "#FF9800", "Faible" = "#FFC107", "Très faible" = "#4CAF50"),
        type = "bar") %>%
  layout(
    title = "Risque de churning par segment",
    xaxis = list(title = ""),
    yaxis = list(title = "Pourcentage de clients"),
    barmode = "stack",
    showlegend = TRUE
  )
```

Analyse Produits {data-icon="fa-box"}
=======================================================================

Row {data-height=100}
-----------------------------------------------------------------------

### {.valueBox}
```{r}
# Nombre total de produits uniques
total_products <- n_distinct(online_retail_clean$StockCode)

valueBox(
  formatC(total_products, format = "d", big.mark = " "),
  caption = "Produits uniques",
  icon = "fa-box",
  color = "primary"
)
```

### {.valueBox}
```{r}
# Produit le plus vendu en quantité
top_product_qty <- product_analysis %>%
  arrange(desc(Quantity)) %>%
  slice(1)

# S'assurer que la description est une chaîne de caractères de longueur 1
product_desc <- if(nrow(top_product_qty) > 0 && !is.na(top_product_qty$Description[1])) {
  substr(as.character(top_product_qty$Description[1]), 1, 20)
} else {
  "Non disponible"
}

valueBox(
  value = product_desc,
  caption = "Produit le plus vendu (quantité)",
  icon = "fa-star",
  color = "success"
)
```

### {.valueBox}
```{r}
# Produit avec le plus haut chiffre d'affaires
top_product_revenue <- product_analysis %>%
  arrange(desc(TotalSales)) %>%
  slice(1)

# S'assurer que la description est une chaîne de caractères de longueur 1
product_revenue_desc <- if(nrow(top_product_revenue) > 0 && !is.na(top_product_revenue$Description[1])) {
  substr(as.character(top_product_revenue$Description[1]), 1, 20)
} else {
  "Non disponible"
}

valueBox(
  value = product_revenue_desc,
  caption = "Produit le plus rentable",
  icon = "fa-euro-sign",
  color = "info"
)
```

### {.valueBox}
```{r}
# Nombre moyen de produits par transaction
avg_products_per_transaction <- online_retail_clean %>%
  group_by(InvoiceNo) %>%
  summarise(ProductCount = n_distinct(StockCode)) %>%
  pull(ProductCount) %>%
  mean()

valueBox(
  formatC(round(avg_products_per_transaction, 1), format = "f", digits = 1),
  caption = "Produits par transaction",
  icon = "fa-shopping-basket",
  color = "warning"
)
```

Row {data-height=500}
-----------------------------------------------------------------------

### Top 10 des produits par chiffre d'affaires
```{r}
# Top 10 des produits par chiffre d'affaires
top10_revenue <- product_analysis %>%
  arrange(desc(TotalSales)) %>%
  head(10)

plot_ly(top10_revenue, 
        x = ~TotalSales, 
        y = ~reorder(substr(Description, 1, 30), TotalSales),
        type = "bar",
        orientation = "h",
        marker = list(color = colorRampPalette(brewer.pal(9, "Blues"))(10))) %>%
  layout(
    title = "Top 10 des produits par chiffre d'affaires",
    xaxis = list(title = "Chiffre d'affaires (€)", tickformat = ",.0f"),
    yaxis = list(title = "")
  )
```

### Top 10 des produits par popularité (nombre de clients)
```{r}
# Top 10 des produits par nombre de clients
top10_customers <- product_analysis %>%
  arrange(desc(CustomerCount)) %>%
  head(10)

plot_ly(top10_customers, 
        x = ~CustomerCount, 
        y = ~reorder(substr(Description, 1, 30), CustomerCount),
        type = "bar",
        orientation = "h",
        marker = list(color = colorRampPalette(brewer.pal(9, "Greens"))(10))) %>%
  layout(
    title = "Top 10 des produits par nombre de clients",
    xaxis = list(title = "Nombre de clients"),
    yaxis = list(title = "")
  )
```

Row {data-height=400}
-----------------------------------------------------------------------

### Produits les plus populaires par segment
```{r}
# Produits les plus populaires par segment
top_products_by_segment <- online_retail_clean %>%
  left_join(rfm_with_lifecycle %>% select(CustomerID, Segment), by = "CustomerID") %>%
  filter(!is.na(Segment)) %>%
  group_by(Segment, StockCode, Description) %>%
  summarise(
    ClientCount = n_distinct(CustomerID),
    TotalSales = sum(TotalPrice)
  ) %>%
  arrange(Segment, desc(ClientCount)) %>%
  group_by(Segment) %>%
  slice(1:3) %>%
  ungroup()

datatable(
  top_products_by_segment %>%
    select(Segment, Description, ClientCount, TotalSales) %>%
    rename(
      "Segment" = Segment,
      "Produit" = Description,
      "Nombre de clients" = ClientCount,
      "Chiffre d'affaires" = TotalSales
    ),
  options = list(
    pageLength = 15,
    order = list(list(0, 'asc'), list(2, 'desc')),
    dom = 'tpl'
  ),
  rownames = FALSE
) %>%
  formatCurrency("Chiffre d'affaires", currency = "€", digits = 2)
```

### Associations de produits (panier moyen)
```{r}
# Préparer les données pour les règles d'association
transactions_for_rules <- online_retail_clean %>%
  select(InvoiceNo, Description) %>%
  unique() %>%
  na.omit()

# Convertir en liste de transactions
transactions_list <- split(transactions_for_rules$Description, transactions_for_rules$InvoiceNo)
transactions <- as(transactions_list, "transactions")

# Extraire les règles d'association
rules <- apriori(
  transactions,
  parameter = list(
    support = 0.005,
    confidence = 0.5,
    minlen = 2,
    maxlen = 2
  )
)

# Convertir en dataframe pour affichage
top_rules_df <- DATAFRAME(sort(rules, by = "lift")[1:10]) %>%
  mutate(
    LHS = gsub("\\{|\\}", "", LHS),
    RHS = gsub("\\{|\\}", "", RHS)
  )

datatable(
  top_rules_df %>%
    select(LHS, RHS, support, confidence, lift) %>%
    rename(
      "Produit A" = LHS,
      "Produit B" = RHS,
      "Support" = support,
      "Confiance" = confidence,
      "Lift" = lift
    ),
  options = list(
    pageLength = 10,
    dom = 'tpl'
  ),
  rownames = FALSE
) %>%
  formatRound(c("Support", "Confiance", "Lift"), digits = 3)
```

Recommandations Personnalisées {data-icon="fa-bullseye"}
=======================================================================

Row {data-height=150}
-----------------------------------------------------------------------

### {.no-padding}
```{r}
# Créer un texte HTML pour la section de recommandations
html_content <- '
<div class="panel panel-primary">
  <div class="panel-heading">
    <h3 class="panel-title" style="text-align: center; font-size: 24px;">
      <i class="fa fa-lightbulb"></i> Recommandations Stratégiques par Segment
    </h3>
  </div>
  <div class="panel-body" style="text-align: center;">
    <p style="font-size: 16px;">Sélectionnez un segment ci-dessous pour voir les recommandations spécifiques et les KPIs à suivre.</p>
  </div>
</div>
'
HTML(html_content)
```

Row {data-height=850}
-----------------------------------------------------------------------

### Recommandations par segment {.tabset .tabset-fade}

```{r segment-recommendations}
# Créer des DataFrames avec les recommandations par segment
champions_recs <- data.frame(
  Action = c("Lancer un programme de fidélité VIP", "Offrir un accès anticipé aux nouveautés", 
             "Envoyer des remerciements personnalisés", "Proposer des offres exclusives",
             "Programme d'ambassadeur", "Événements VIP", "Conciergerie dédiée"),
  Type = c("Immédiate", "Immédiate", "Immédiate", "Immédiate", 
           "Moyen terme", "Moyen terme", "Moyen terme"),
  stringsAsFactors = FALSE
)

loyals_recs <- data.frame(
  Action = c("Programmes de vente incitative", "Programme de fidélité avec paliers", 
             "Offres personnalisées", "Récompenser l'engagement régulier",
             "Programme de parrainage", "Contenus exclusifs", "Stratégies de vente croisée"),
  Type = c("Immédiate", "Immédiate", "Immédiate", "Immédiate", 
           "Moyen terme", "Moyen terme", "Moyen terme"),
  stringsAsFactors = FALSE
)

potential_recs <- data.frame(
  Action = c("Offres pour augmenter la fréquence", "Recommandations personnalisées", 
             "Promotions produits déjà achetés", "Programme d'engagement par email",
             "Parcours d'achat personnalisés", "Contenu éducatif", "Système de rappel d'achat"),
  Type = c("Immédiate", "Immédiate", "Immédiate", "Immédiate", 
           "Moyen terme", "Moyen terme", "Moyen terme"),
  stringsAsFactors = FALSE
)

atrisk_recs <- data.frame(
  Action = c("Campagne de réactivation", "Enquête de satisfaction", 
             "Remise exceptionnelle", "Contact téléphonique (clients haute valeur)",
             "Identifier points de friction", "Programme de reconquête", "Système d'alerte précoce"),
  Type = c("Immédiate", "Immédiate", "Immédiate", "Immédiate", 
           "Moyen terme", "Moyen terme", "Moyen terme"),
  stringsAsFactors = FALSE
)

dormant_recs <- data.frame(
  Action = c("Campagne 'Dernière chance'", "Remise substantielle", 
             "Enquête raisons de départ", "Programme 'bienvenue à nouveau'",
             "Nouvelle proposition de valeur", "Segment clients reconquis", "Nettoyage base de données"),
  Type = c("Immédiate", "Immédiate", "Immédiate", "Immédiate", 
           "Moyen terme", "Moyen terme", "Moyen terme"),
  stringsAsFactors = FALSE
)
```

#### Champions

```{r}
# Premier point d'information avec couleur de fond
div(class = "alert alert-success",
    h4(tags$i(class = "fa fa-trophy"), " Champions"),
    p(strong("Description : "), "Clients récents à haute valeur et fréquence d'achat élevée. Ces clients sont vos meilleurs ambassadeurs.")
)

# Afficher les recommandations dans un tableau
datatable(
  champions_recs,
  options = list(
    pageLength = 10,
    dom = 't',
    ordering = FALSE
  ),
  rownames = FALSE,
  caption = htmltools::tags$caption(
    style = "caption-side: top; text-align: center; font-size: 16px; margin-bottom: 10px;",
    htmltools::tags$strong("Actions recommandées")
  )
)

# KPIs à suivre
div(class = "row",
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Rétention")),
            tags$ul(
              tags$li("Taux de rétention (>90%)"),
              tags$li("Durée de vie client (LTV)"),
              tags$li("Taux d'attrition")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Engagement")),
            tags$ul(
              tags$li("Taux d'adoption offres VIP"),
              tags$li("Taux d'ouverture emails"),
              tags$li("Net Promoter Score")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Valeur")),
            tags$ul(
              tags$li("Valeur panier moyen"),
              tags$li("Fréquence d'achat"),
              tags$li("Conversion ventes incitatives")
            )
        )
    )
)
```

#### Clients Fidèles

```{r}
div(class = "alert alert-info",
    h4(tags$i(class = "fa fa-heart"), " Clients Fidèles"),
    p(strong("Description : "), "Clients réguliers avec une valeur modérée à élevée. Fidèles à votre marque avec potentiel d'évolution.")
)

# Afficher les recommandations dans un tableau
datatable(
  loyals_recs,
  options = list(
    pageLength = 10,
    dom = 't',
    ordering = FALSE
  ),
  rownames = FALSE,
  caption = htmltools::tags$caption(
    style = "caption-side: top; text-align: center; font-size: 16px; margin-bottom: 10px;",
    htmltools::tags$strong("Actions recommandées")
  )
)

# KPIs à suivre
div(class = "row",
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Rétention")),
            tags$ul(
              tags$li("Taux de rétention (>80%)"),
              tags$li("Taux de réachat"),
              tags$li("Intervalle entre achats")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Croissance")),
            tags$ul(
              tags$li("Migration vers segment Champions"),
              tags$li("Augmentation panier moyen"),
              tags$li("Conversion ventes croisées")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Engagement")),
            tags$ul(
              tags$li("Participation programme fidélité"),
              tags$li("Nombre de recommandations"),
              tags$li("Engagement communications")
            )
        )
    )
)
```

#### Clients Potentiels

```{r}
div(class = "alert alert-warning",
    h4(tags$i(class = "fa fa-seedling"), " Clients Potentiels"),
    p(strong("Description : "), "Clients récents avec fréquence et valeur moyennes. Potentiel de devenir clients fidèles.")
)

# Afficher les recommandations dans un tableau
datatable(
  potential_recs,
  options = list(
    pageLength = 10,
    dom = 't',
    ordering = FALSE
  ),
  rownames = FALSE,
  caption = htmltools::tags$caption(
    style = "caption-side: top; text-align: center; font-size: 16px; margin-bottom: 10px;",
    htmltools::tags$strong("Actions recommandées")
  )
)

# KPIs à suivre
div(class = "row",
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Activation")),
            tags$ul(
              tags$li("Conversion des offres"),
              tags$li("Augmentation fréquence"),
              tags$li("Adoption programme fidélité")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Engagement")),
            tags$ul(
              tags$li("Taux d'ouverture emails"),
              tags$li("Engagement site web"),
              tags$li("Nombre d'avis laissés")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Progression")),
            tags$ul(
              tags$li("Migration segments supérieurs"),
              tags$li("Augmentation valeur panier"),
              tags$li("Diversité produits achetés")
            )
        )
    )
)
```

#### Clients à Risque

```{r}
div(class = "alert alert-danger",
    h4(tags$i(class = "fa fa-exclamation-triangle"), " Clients à Risque"),
    p(strong("Description : "), "Clients anciennement actifs sans achat récent. Risque élevé de churning sans réactivation.")
)

# Afficher les recommandations dans un tableau
datatable(
  atrisk_recs,
  options = list(
    pageLength = 10,
    dom = 't',
    ordering = FALSE
  ),
  rownames = FALSE,
  caption = htmltools::tags$caption(
    style = "caption-side: top; text-align: center; font-size: 16px; margin-bottom: 10px;",
    htmltools::tags$strong("Actions recommandées")
  )
)

# KPIs à suivre
div(class = "row",
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Réactivation")),
            tags$ul(
              tags$li("Taux de réactivation"),
              tags$li("Réponse aux campagnes"),
              tags$li("Conversion offres spéciales")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Satisfaction")),
            tags$ul(
              tags$li("Réponse aux enquêtes"),
              tags$li("Score de satisfaction"),
              tags$li("Points de friction identifiés")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Rétention")),
            tags$ul(
              tags$li("Taux de churn"),
              tags$li("Valeur clients réactivés"),
              tags$li("Fréquence après réactivation")
            )
        )
    )
)
```

#### Clients Dormants

```{r}
div(class = "alert alert-secondary",
    h4(tags$i(class = "fa fa-bed"), " Clients Dormants"),
    p(strong("Description : "), "Clients inactifs depuis longtemps avec faible valeur et fréquence. Possibilité de reconquête avec bonne approche.")
)

# Afficher les recommandations dans un tableau
datatable(
  dormant_recs,
  options = list(
    pageLength = 10,
    dom = 't',
    ordering = FALSE
  ),
  rownames = FALSE,
  caption = htmltools::tags$caption(
    style = "caption-side: top; text-align: center; font-size: 16px; margin-bottom: 10px;",
    htmltools::tags$strong("Actions recommandées")
  )
)

# KPIs à suivre
div(class = "row",
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Reconquête")),
            tags$ul(
              tags$li("Taux de reconquête"),
              tags$li("ROI campagnes reconquête"),
              tags$li("Coût acquisition vs reconquête")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Feedback")),
            tags$ul(
              tags$li("Réponse aux enquêtes"),
              tags$li("Raisons principales d'inactivité"),
              tags$li("Suggestions d'amélioration")
            )
        )
    ),
    div(class = "col-md-4",
        div(class = "well",
            h5(strong("KPIs Optimisation")),
            tags$ul(
              tags$li("Taux de désabonnement"),
              tags$li("Nettoyage de la base"),
              tags$li("Valeur clients réactivés")
            )
        )
    )
)
```